Data-as-a-Service
Data-as-a-Service: Data or data management as a core underlying asset; services are created on top of data that is collected, trained, and labeled
Moving Beyond Insights and Analytics
Much of data strategy and delivery has been constrained and siloed within the Business Intelligence function inside of companies. Data-as-a-service models move beyond merely enhancing data processes and analytics to inform internal decisions and move to create value for end customers outside the organizations.
Data-as-a-Service vs. Artificial Intelligence / Machine Learning
AI and ML are data science and computational processes that can extend to any digital business model. Data-as-a-Service is a distinct business model approach that seeks to place a value on a data process that customers will pay for.
Business Models in Use
Blockscore | Bloomberg | Climate Corp (acquired by Monsanto) | Epsilon | Experian | Factual | Intelius | IOTA | LiveRamp | People Lookup | PIPL | Qlik | Sift Science | Socrata | Streamr | ZabraSearch | Zephyr Health
Data-as-a-Service Evaluator
Valuable to Customer
Benefits
- Value of data outside the firm greater than proprietary data inside
- Opportunities to enhance revenue streams when built into key workflows
- Data enhances work vs. replaces through automation
Challenges
- Risk managing security
- Governance and privacy concerns when integrating with customer data or managing anonymized datasets
Key Performance Indicators
- ROI data value vs. cost
- Size and volume of us
COMPANY PERSPECTIVE
Benefits
- Huge potential value if DaaS company has hard-to-access data
- Value in structured, cleaned or scrubbed data
- Value of data can be the primary driver of customer and company value
Challenges
- Security issues (i.e. Equifax)
- Regulatory and user privacy concerns
- Cost of data operations not always in line with the value of data solution
Key Performance Indicators
- Data acquisition cost, volume, velocity, variety
- Net user growth accessing data types
- Volumes of data use
- Data usage and user retention
When it Works Well
Fills Critical Data and Data Quality Gaps
Business data used to be generated primarily from inside of a company, and from interaction directly with customers. Data sources were static, structured, and small enough to be stored or digitally warehoused inside of a company (often in physical servers).
Now, most data lives outside of a company’s walls, the data is big, which means there is too much, too fast, with too much variety to be stored in-house. Consumers, sensors, and machines – most of them outside the walls of the enterprise – are now generating most of the data in the world. The data has a different value and different uses, and the data should do the work of data preparation: fill data gaps, filter, clean, aggregate, and simplify data access for the customer.
Companies can generate their own dataset and sell access (Factual), build one of the most effective access points and sell access (Bloomberg), or help customers add external data to their datasets (Zephyr Health or Socrata).
Feed a Data Chain that Drives Revenue or Cost Reduction
Business data used to be generated primarily from inside of a company, and from interaction directly with customers. Data sources were static, structured, and small enough to be stored or digitally warehoused inside of a company (often in physical servers).
Now, most data lives outside of a company’s walls, the data is big, which means there is too much, too fast, with too much variety to be stored in-house. Consumers, sensors, and machines – most of them outside the walls of the enterprise – are now generating most of the data in the world. The data has a different value and different uses, and the data should do the work of data preparation: fill data gaps, filter, clean, aggregate, and simplify data access for the customer.
Companies can generate their own dataset and sell access (Factual), build one of the most effective access points and sell access (Bloomberg), or help customers add external data to their datasets (Zephyr Health or Socrata).
Integrates Into a Workflow
Data-as-a-service models typically do not disrupt a core process but instead augment a workflow. Customer experience journeys and a deep understanding of the industry needs help to immediately provide value to the business. Few customers ask for an additional dashboard, so data services companies often follow a default design strategy to integrate into the day of the end-user, and a well-defined customer problem.
Builds Moats Around the Data
As software and hardware are commoditized, deriving value from meaningful datasets remains a core source of competitive advantage. While many leading software-driven companies open source their management, infrastructure, and monitoring tools; they tend to keep their data as a primary asset. Because it is difficult to copy a dataset without a large user base, companies focus on building a moat around the data.
Challenges to the DaaS Model
Margins Uglier in the Beginning
Because of the way that data costs are accounted for early on investments to acquire data result in high cost of goods sold (COGS) and therefore high margins. However, these margins do indeed increase over time. But investors used to seeing SaaS margins may overlook an opportunity to invest in a data services company at the start.
Data Exhaust and Surveillance Capitalism Concerns
Because tech giants have been less than transparent about how they collect, store, barter, and sell data there is emerging concern that these companies are undermining personal autonomy and eroding democracy. We would argue that the business model matters. Companies that collect data exhaust or data sources that use advertising as their primary business model operate very differently than companies with subscription-based or more diverse business models.
Trends in DaaS
Self-Service Insights
Data-driven companies are going through a renaissance of experimentation, as new startups and existing firms experiment with new data business models. Many new firms are moving upstream to specialize in data visualization, data-driven insights, and decision tools, providing directly to business analysts and decision-makers, and breaking down the central hierarchy of the CTO, CIO, and information technology staff. Companies are also designing end customer-facing data-driven insights with the advent of sensors in everything: golf clubs that inform you how to improve your game, or sensor tags helping you find your keys.
Data Commons
There are emerging industry, foundation, not-for-profit, cooperative, and non-governmental organization solutions to collect and manage data to protect privacy while also making data available for public good research.
Automated Insights
As insights move upstream, the connection of everything to everybody means that previously invisible processes can now be tracked and optimized. Everything that can be automated will be automated, replacing analog and human processes. As Amazon’s CFO Brian Olsavsky said in the Second Quarter 2015 financial report, “We’re using software and algorithms to make decisions rather than people, which we think is more efficient and scales better.” Uber is currently looking for engineers who are excited by such as prospect: “Do you enjoy having your software make critical decisions rather than just “process” or “display”?
Diminishing Returns for Once Valuable Proprietary Datasets
Data collection and supply is not a new business. Experian and Axciom are public companies that have been amassing large data stores to sell to companies, and these firms are infamous for the decisions they inform: ad targeting, mortgage rates, credit card terms, insurance premiums, and employment offers.
But the rise in the huge volume of new data sources and varieties diminishes the value of data collected in these proprietary data gatherers. As business model innovation continues at an accelerated rate, expect traditional business metrics and the dominant logic of the industry to shift.
Customers See Value in Data and Want to Complete Ownership
You may have designs on a future data asset to power a massive data service, but your customers may not want you to collect that data. Developing agreements based on anonymized data, or developing revenue share models may help you to mitigate this challenge.
Similarly, end users are now organizing to get access to their data and advocate for full ownership.
New Privacy and Data Ownership Regulation
Governments at the national and state level have put specific data ownership and privacy standards in place. Following these regulations to their highest standards has become costly for startups, and often benefits incumbents. The General Data Protection Regulation (GDPR) implemented in 2018 has higher standards for any company collecting user data. New opportunities emerge for more open published standards on data use, and for new entrants that address user ownership of data as a fundamental core assumption within the business model.
Before You Consider DaaS
- Where does the solution fall on the data-knowledge-wisdom hierarchy?
- Test for jobs to be done, level, pain on the pain scale. How much of a priority is this solution?
- Is your data ready to be used? Do we require additional data from the customer, third parties, or open-source data projects?Is there a true competitive advantage to your hardware; can it be replaced by a SaaS offering
- What is the culture of data-informed decisions within our customer base?
- Are our customers data-literate? Can they review Python libraries or dashboards? Do they need to review everything in Powerpoint with an analyst at the ready?
- What decisions will be affected by our data solution?
Testing the Model
- What is the total cost of ownership of comparable solutions?
- Arrange features, services, and benefits into key elements of your offer and have the potential customer arrange the elements of the larger solution in order of priority. Then take away the lesser priority elements until you determine what would make an MVP (minimum viable product).
- Determine the minimal offering that would be compelling enough to have the customer pay for the offering.
- Can you design an MVP that has high usage and engagement with a minimal feature set?
- Is there a user proposition that does not require sign-off from IT or a long buying cycle?
More on DaaS
Value of Data: Using Data as a Competitive Advantage, by Lee Polovets at Susa Ventures, 2015.
Data Driven Business Models, Cambridge Service Alliance, 2014. (PDF)
Data Markets and the Monetisation of Data Goods, Intereconomics, 2019
Data-As-A-Service Bible: Everything You Wanted To Know About Running DaaS Companies, by Auren Hoffman, Safegraph, 2019.
Never Sell Data, by Steve Faktor, Forbes.com, 2014.
AI business model: an integrative business approach, by Sukrita Mishra, Journal of Innovation and Entrepreneurship, 2021.
The Attributes that Define the Increasingly Critical Data-as-a-Service Industry, by Alex Rosen, TechCrunch, 2018.
A Review of the Data Broker Industry: Collection, Use, and Sale of Consumer Data for Marketing Purposes, Staff Report for Chairman Rockefeller, Committee on Commerce, Science and Transportation, US Senate, 2013 (PDF)
The Morning Download Amazon CFO Says Algorithm-Based Decision Making Helped Company Achieve Profit, by Steve Rosenbush, The Wall Street Journal, 2015.
High tech is watching you, by Shoshana Zuboff, NYT, 2019.
Personal Data as an Asset Class, World Economic Forum, 2011. (PDF)
Data is the New Crude, by Michael Palmer, ANA Maestros, 2006.
Data-as-a-Service, by Rocketsource, 2017
Big Data is Not the New Oil, by Jer Thorpe, Harvard Business Review, 2012. (limited access)
Business Models for the Data Economy, by Q Ethan McCallum and Ken Gleason, O’Reilly Media, 2013 (download)
Everything We Wish We’d Known about Building Data Products, First Round Review, 2015.
Data As A Service: The Big Opportunity For Business, by Daniel Newman, Forbes, 2017
